R-matrix equality makes the comparison endpoints exact
Let \(\phi:[a,b]\simeq[a',b']\), and assume equality of the normalized R-polynomials for every corresponding subinterval: \[ \widetilde R_{x,z}=\widetilde R'_{\phi(x),\phi(z)} \qquad(a\leq x\leq z\leq b). \] For the actual boundary sheaf restricted to \([a,b]\), the schedule comparison has equal endpoints for every \(0<t<1\).
Relation to the OpenAI notes
Different proof route.
OpenAI notes: The simultaneous induction compares normalized KL vectors, obtains P-invariance from the strict half-degree bound, and then recovers bottom terminal-kernel freeness from endpoint equality. sec:induction, eq:comparison, eq:reciprocal-vector.
Lean (primary): The simultaneous induction first proves normalized R-invariance together with suffix-kernel freeness in both transported directions. It cancels known bottom R-row entries against one actual source-character vector. Canonical KL equality is proved afterwards in a separate induction using the now common R-matrix and the canonical degree bound. Invariance:136, Invariance:190, Invariance:427.
Role in the proof: Proper upper intervals supply kernel freeness, and proper initial intervals supply known R-entries. The new R-equality gives exact endpoints and then current-rank suffix freeness. The canonical degree bound enters the subsequent P-induction.
Proof
Step 1. Identify the initial and final states
Let \(p\) be the full chronological schedule and let \(c_x(t)\) be the scaled actual source stalk character. The empty schedule uses the whole stalk, while the full schedule uses its full outgoing kernel. With \(q=t^2\), their Hilbert values are \(C_{x,b}(q)/(1-q)^N\) and \(q^{d(x,b)}C_{x,b}(q^{-1})/(1-q)^N\), respectively, by The grading shift in stalk–costalk reciprocity. Multiplying by the scale \((1-t^2)^Nt^{-d(x,b)}\) cancels the common denominator, and the exponents become \(-d(x,b)\) and \(d(x,b)\). Hence \[ D_{[]}(t)=c(t),\qquad D_p(t)=c(t^{-1}). \] The full list contains every outgoing condition; the empty list contains none. This is the precise reason those two character vectors occur.
Step 2. Expand the transported product
For \(T=t^{-1}-t\), the ordered-path formula gives, coordinatewise, \[ \left(\operatorname{Dispatch}_{p^{\mathrm{rev}}}(T)D_p(t)\right)_x =\sum_{z\in[a,b]} \widetilde R'_{\phi(x),\phi(z)}(T)c_z(t^{-1}). \] The decreasing schedule is \(p^{\mathrm{rev}}\). The root labels used to define its scores belong to the target system, which explains the prime on the R-polynomials.
Step 3. Use all-entry R equality
For comparable \(x\le z\), substitute the assumed equality \(\widetilde R'_{\phi(x),\phi(z)}=\widetilde R_{x,z}\). For noncomparable pairs, both polynomials vanish, since \(\phi\) preserves the order. Thus every summand becomes its source-system counterpart: \[ \sum_z\widetilde R_{x,z}(T)c_z(t^{-1}). \]
Step 4. Apply the source reciprocity identity
The scaled actual-character reciprocity theorem states that this last sum is \(c_x(t)\). By Step 1, it is \(D_{[]}(t)_x\). This holds for each \(x\), so the two vectors are equal: \[ D_{[]}(t)=\operatorname{Dispatch}_{p^{\mathrm{rev}}}(T)D_p(t). \] This completes the coordinatewise endpoint identity.
In the simultaneous induction, the smaller subintervals supply the other comparable R-entries, and the newly proved current entry completes the matrix equality used here. Choosing \(t=\tfrac12\) gives one concrete exact endpoint, sufficient for the later Hilbert-equality argument.
References
Lean formalization
OAI.KLInvariance.BruhatGraph.actual_endpoint_of_rTilde_eq: SimultaneousInvariance.lean:278.